Agentscore

@agentscore-xyz/mcp-server

MCP security trust layer. Scan packages, get trust verdicts, inspect repo-wide MCP dependencies, generate Policy Gate setup, install the CI workflow directly, check incident exposure, and query the abuse database. Eight tools for MCP security decisions. No API key, zero config.

Scan any MCP package for security issues: agentscores.xyz

Quick Start

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "agentscore": {
      "command": "npx",
      "args": ["-y", "@agentscore-xyz/mcp-server"]
    }
  }
}

Cursor / Any MCP Client

npx @agentscore-xyz/mcp-server

What This Does

Your AI can now make security decisions about MCP packages:

You: "Is exa-mcp-server safe to install?"

Claude: calls get_verdict "Verdict: ALLOW. Score 90/100, LOW risk. No provenance attestations (published by personal account). 9 tools exposed including web_search_exa and crawling_exa."

You: "The axios package was compromised. Which MCP servers are affected?"

Claude: calls check_exposure "Multiple monitored MCP servers depend on axios, including exa-mcp-server, tavily-mcp, and figma-mcp."

You: "Scan @azure-devops/mcp for security issues"

Claude: calls scan_package "Score 75/100, MODERATE risk. Found: preinstall script modifying npm registry config. No provenance attestations."

You: "Check this repo for MCP dependencies"

Claude: calls check_my_repo "MCP dependencies found: 5. Two are warnings. Run generate_policy_gate_setup to turn these checks into a CI gate."

You: "Set up AgentScore Policy Gate for this repo"

Claude: calls install_policy_gate "The workflow file is written to .github/workflows/agentscore-policy-gate.yml. Commit and push. GitHub OIDC will auto-provision the repo on first run."

Available Tools

Tool	What it does
scan_package	Full security scan: install scripts, prompt injection, source code patterns, provenance posture, MCP tool extraction
get_verdict	Trust decision: allow, warn, or block based on scan findings. Also reports monitoring status and publisher posture.
check_my_repo	Inspect the current repo for MCP dependencies and summarise verdicts for every package detected locally.
generate_policy_gate_setup	Generate the exact OIDC-based GitHub Actions workflow needed to enforce Policy Gate in CI.
install_policy_gate	Write .github/workflows/agentscore-policy-gate.yml directly into the repo so the gate is ready to commit.
check_exposure	Incident response: which monitored MCP servers depend on a given package?
check_abuse	Query the KYA abuse database for reported packages or agents
monitor_status	Check if a package is under continuous monitoring and get scan history

From Ad-Hoc Scans To CI Enforcement

The MCP server now bridges one-off package checks into the sticky product:

1. Run check_my_repo to see every MCP package used in a repo.
2. Run generate_policy_gate_setup to preview the OIDC-based GitHub Actions workflow.
3. Run install_policy_gate to write the workflow file directly into the repo.
4. Commit and push. The first run auto-provisions through GitHub OIDC.

That turns "is this package safe?" into "this repo now enforces MCP dependency policy on every PR."

Risk Levels

Score	Risk	Meaning
85-100	LOW	Clean or minor issues only
70-84	MODERATE	Some findings, review recommended
50-69	ELEVATED	Significant findings, use with caution
30-49	HIGH	Serious issues, not recommended
0-29	CRITICAL	Do not use

What the Scanner Checks

• Install scripts (postinstall/preinstall hooks with network calls or code execution)
• Prompt injection patterns in package metadata
• Suspicious URLs (sketchy TLDs, ngrok, raw IPs)
• Source code patterns (command injection, unsafe eval, hardcoded secrets)
• Publisher provenance (trusted publishing, attestations)
• Dependency count and metadata completeness
• MCP tool definitions extracted from published source

Monitoring

AgentScore continuously monitors hundreds of MCP packages. The check_exposure and monitor_status tools use this live dataset. When a package like axios gets compromised, you can instantly find which MCP servers are affected.

Links

• Website: agentscores.xyz
• API Docs: agentscores.xyz/docs
• Methodology: agentscores.xyz/methodology

Licence

MIT